The quality of processes and products can be determined by analyzing operating and functional noises. For example, machine operators often notice a defect in a system by its changed operating noise. In order to automate this process, acoustic evaluation methods are required which can make a well-founded iO/niO decision. According to the current state of the art, such methods are primarily based on body and ultrasonic analyses. There are hardly any established methods for sound evaluation in the audible frequency spectrum. This is due to various challenges associated with the evaluation of audible sounds. On the one hand, existing background noises must not be included in the evaluation and all aspects of data protection and data security must be taken into account. A robust solution for non-contact acoustic quality assurance requires some prerequisites and intelligent methods of machine learning. Previous industry examples - the scenarios come from the areas of predictive maintenance and end-of-line testing - show both the performance and the limits of the technology. In addition, the article gives an outlook on further main research areas resulting from the application of the above-mentioned technologies and the further development of these processes.